Practice with dplyr and ggplot2

Fancy Histograms!

We’ll start by making some histograms.

#install.packages("dslabs")
library(dslabs) 
data(heights)
glimpse(heights)

## Rows: 1,050
## Columns: 2
## $ sex    <fct> Male, Male, Male, Male, Male, Female, Female, Female, Female, M…
## $ height <dbl> 75, 70, 68, 74, 61, 65, 66, 62, 66, 67, 72, 72, 69, 68, 69, 66,…

This data is the heights of humans, divided by their biological sex.

Use ggplot to make a histogram of all of the heights:

ggplot(heights, aes(x = height)) + geom_histogram()

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

Change up the binwidth and see how the plots change. Try 1, 5, 10, and 20

ggplot(heights, aes(x = height)) + geom_histogram(binwidth = 1)

Smooth this out to an emperical density with geom_density()

ggplot(heights, aes(x = height)) + geom_density()

Use a new argument in the aes(), group = to split this density by sex

ggplot(heights, aes(x = height, group = sex)) + geom_density()

OR we can do it with color or fill. If you say you want to color by sex, R knows that you want a different curve for each of them.

ggplot(heights, aes(x = height, color = sex)) + geom_density()

If you’ve used fill, then there is now a slight issue that they are overlapped. We can fix this with alpha transparency!

ggplot(heights, aes(x = height, fill = sex)) + geom_density(alpha = .3)

Let’s make some boxplots of the same information.

ggplot(heights, aes(x = height, y = sex, fill = sex)) + geom_boxplot()

Quantatitive summaries:

Find the mean and median overall.

heights %>% summarise(overallmean = mean(height), overallmedian = median(height))

##   overallmean overallmedian
## 1    68.32301          68.5

Find the mean and median for both groups.

heights %>% group_by(sex) %>% summarise(mean = mean(height), median = median(height))

## # A tibble: 2 × 3
##   sex     mean median
##   <fct>  <dbl>  <dbl>
## 1 Female  64.9   65.0
## 2 Male    69.3   69

How tall is the tallest woman? How short is the shortest man?

heights %>% group_by(sex) %>% summarise(tallest = max(height), shortest = min(height))

## # A tibble: 2 × 3
##   sex    tallest shortest
##   <fct>    <dbl>    <dbl>
## 1 Female    79         51
## 2 Male      82.7       50

Presidental Elections Data

# install.packages("pscl")
library(pscl) # loads in the package that has this data. 

## Classes and Methods for R originally developed in the
## Political Science Computational Laboratory
## Department of Political Science
## Stanford University (2002-2015),
## by and under the direction of Simon Jackman.
## hurdle and zeroinfl functions by Achim Zeileis.

## You might need to install this...

# data for presidental elections
votedata <-  presidentialElections
glimpse(votedata)

## Rows: 1,097
## Columns: 4
## $ state   <chr> "Alabama", "Arizona", "Arkansas", "California", "Colorado", "C…
## $ demVote <dbl> 84.76, 67.03, 86.27, 58.41, 54.81, 47.40, 48.11, 74.49, 91.60,…
## $ year    <int> 1932, 1932, 1932, 1932, 1932, 1932, 1932, 1932, 1932, 1932, 19…
## $ south   <lgl> TRUE, FALSE, TRUE, FALSE, FALSE, FALSE, FALSE, TRUE, TRUE, FAL…

Let’s look at the democratic vote by state for 2000. We can’t use geom_bar for a bar chart, since we have the category in one variable and the “height” of the bar in another. We need geom_col()

Make a bar graph of the democratic vote by state in 2000.

votedata %>% filter(year == "2000") %>% ggplot(aes(x = state, y = demVote)) + geom_col()

Well this looks awful. We have two options: swap the x and y or the more fun sounding… Coordinate flip!

Use coord_flip() on the previous graph to make it better.

votedata %>% filter(year == "2000") %>% ggplot(aes(x = state, y = demVote)) + geom_col() + coord_flip()

I don’t love the squashed together coordinates, but it’s a display window issue.

So. This is a helpful graph, but it would be more helpful if it was ordered. Use x = reorder(x_variable, y_variable) in aes() to order the x variable by the y variable

votedata %>% filter(year == "2000") %>% ggplot(aes(x = reorder(state, demVote), y = demVote)) + geom_col() + coord_flip()

So, what if I want to see what the north and south states did different?

start with a facet_wrap using the south variable:

votedata %>% filter(year == "2000") %>% ggplot(aes(x = reorder(state, demVote), y = demVote)) + geom_col() + coord_flip() + facet_wrap(vars(as.factor(south)))

Okay, that’s not great. Lets color by south instead.

votedata %>% filter(year == "2000") %>% ggplot(aes(x = reorder(state, demVote), y = demVote, fill = south)) + geom_col() + coord_flip() + scale_fill_manual(values = c("TRUE" = "#94b0da", "FALSE" = "#d7263d"))

I’m a good data scientist, so I want my plot to have a name! and my axes to have lables! Use labs to add a title, subtitle, and x and y labels.

votedata %>% filter(year == "2000") %>% ggplot(aes(x = reorder(state, demVote), y = demVote, fill = south)) + 
  geom_col() + coord_flip() + 
  scale_fill_manual(
    name = "Region",  # Legend title
    values = c("TRUE" = "#94b0da", "FALSE" = "#d7263d"),
    labels = c("TRUE" = "Non-South", "FALSE" = "South")) +
  labs(title = "Percentage of Vote Won by Democratic Candidate", subtitle = "US Presidential Race of 2000", x = "Democrat Vote", y = "State")

You can move the legend with theme(legend.position = "bottom")

votedata %>% filter(year == "2000") %>% ggplot(aes(x = reorder(state, demVote), y = demVote, fill = south)) + 
  geom_col() + coord_flip() + 
  scale_fill_manual(
    name = "Region",  # Legend title
    values = c("FALSE" = "#94b0da", "TRUE" = "#d7263d"),
    labels = c("FALSE" = "Non-South", "TRUE" = "South")) +
  labs(title = "Percentage of Vote Won by Democratic Candidate", subtitle = "US Presidential Race of 2000", y = "State") + theme(legend.position = "bottom")

What else could we facet by? years! Let’s filter to year in 2008 and 2016, then facet by years.

votedata %>% filter(year == "2008" | year == "2016") %>% ggplot(aes(x = reorder(state, demVote), y = demVote, fill = south)) + 
  geom_col() + coord_flip() + 
  scale_fill_manual(
    name = "Region",  # Legend title
    values = c("FALSE" = "#94b0da", "TRUE" = "#d7263d"),
    labels = c("FALSE" = "Non-South", "TRUE" = "South")) +
  labs(title = "Percentage of Vote Won by Democratic Candidate", subtitle = "US Presidential Race", x = "Democrat Vote", y = "State") + theme(legend.position = "bottom") + facet_wrap(~ year)

We need to know who won! We could add a vertical line at 50 for who got more, to indicate the majority of votes. Adding the layer geom_hline() adds a horizontal line. (What do you guess geom_vline() would do?)

• Would add a line in the opposite direction. Even though hline sounds like horizontal, line, since we flipped coordinates we also flipped the hline to be vertical.

votedata %>% filter(year == "2008" | year == "2016") %>% ggplot(aes(x = reorder(state, demVote), y = demVote, fill = south)) + 
  geom_col() + coord_flip() + 
  scale_fill_manual(
    name = "Region",  # Legend title
    values = c("FALSE" = "#94b0da", "TRUE" = "#d7263d"),
    labels = c("FALSE" = "Non-South", "TRUE" = "South")) +
  labs(title = "Percentage of Vote Won by Democratic Candidate", subtitle = "US Presidential Race", x = "Democrat Vote", y = "State") + theme(legend.position = "bottom") + 
  facet_wrap(~ year) +
  geom_hline(yintercept = 50)

Getting fancy with a map!

When using geom_polygon or geom_map, you will typically need two data frames:

• one contains the coordinates of each polygon (positions)
• the other the values associated with each polygon (values).

An id variable links the two together.

Run the below code to get a map graph.

library(maps)

## 
## Attaching package: 'maps'

## The following object is masked from 'package:purrr':
## 
##     map

votedata$state <- tolower(votedata$state)  ## states need to be lowercase for linking

states_map <- map_data("state") ## this gives us the lat and long for each point of each state.
  
map_plot <-  ggplot(data =  votedata %>% filter(year == 2008), aes(map_id = state)) +
    geom_map(aes(fill = demVote), map = states_map) +
    expand_limits(x = states_map$long, y = states_map$lat)
map_plot  

map_plot <-  ggplot(data =  votedata %>% filter(year == 2016), aes(map_id = state)) +
  geom_map(aes(fill = demVote), map = states_map)+
  expand_limits(x = states_map$long, y = states_map$lat)
map_plot  

What if I want a map that shows which of the states are “south”? What do I change?

map_plot <-  ggplot(data =  votedata %>% filter(year == 2008), aes(map_id = state)) +
    geom_map(aes(fill = south), map = states_map) +
    expand_limits(x = states_map$long, y = states_map$lat)
map_plot  

Some more dplyr practice

I want to know the average democratic vote for N vs S, by year.

First, find the average democratic votes for the north and the south, every year. You’ll need to do a double group_by() here. You do it in one call of the function.

votedata %>% group_by(year, south) %>% summarise(averagevotes = mean(demVote))

## `summarise()` has grouped output by 'year'. You can override using the
## `.groups` argument.

## # A tibble: 44 × 3
## # Groups:   year [22]
##     year south averagevotes
##    <int> <lgl>        <dbl>
##  1  1932 FALSE         56.7
##  2  1932 TRUE          83.4
##  3  1936 FALSE         59.2
##  4  1936 TRUE          83.2
##  5  1940 FALSE         52.8
##  6  1940 TRUE          80.9
##  7  1944 FALSE         51.1
##  8  1944 TRUE          75.1
##  9  1948 FALSE         50.2
## 10  1948 TRUE          45.9
## # ℹ 34 more rows

Then, let’s plot that! Pipe the result of your group_by and summarize to ggplot and geom_line(), with year on the x axis and your summarized value on the y axis. Color by the south variable.

votedata %>% group_by(year, south) %>% summarise(averagevotes = mean(demVote)) %>% ggplot(aes(x = year, y = averagevotes, color = south)) + geom_line()

## `summarise()` has grouped output by 'year'. You can override using the
## `.groups` argument.

Layering plots!

Penguins!

library(palmerpenguins)
glimpse(penguins)

## Rows: 344
## Columns: 8
## $ species           <fct> Adelie, Adelie, Adelie, Adelie, Adelie, Adelie, Adel…
## $ island            <fct> Torgersen, Torgersen, Torgersen, Torgersen, Torgerse…
## $ bill_length_mm    <dbl> 39.1, 39.5, 40.3, NA, 36.7, 39.3, 38.9, 39.2, 34.1, …
## $ bill_depth_mm     <dbl> 18.7, 17.4, 18.0, NA, 19.3, 20.6, 17.8, 19.6, 18.1, …
## $ flipper_length_mm <int> 181, 186, 195, NA, 193, 190, 181, 195, 193, 190, 186…
## $ body_mass_g       <int> 3750, 3800, 3250, NA, 3450, 3650, 3625, 4675, 3475, …
## $ sex               <fct> male, female, female, NA, female, male, female, male…
## $ year              <int> 2007, 2007, 2007, 2007, 2007, 2007, 2007, 2007, 2007…

We can use boxplots to visualize the distribution of weight (body_mass_g) within each species:

penguins %>% ggplot(aes(x = species, y = body_mass_g)) + geom_boxplot()

## Warning: Removed 2 rows containing non-finite outside the scale range
## (`stat_boxplot()`).

What if we also want the points? Layering!! Add a geom_point to your existing boxplot. geom_boxplot + geom_point!

penguins %>% ggplot(aes(x = species, y = body_mass_g)) + geom_boxplot() + geom_point()

## Warning: Removed 2 rows containing non-finite outside the scale range
## (`stat_boxplot()`).

## Warning: Removed 2 rows containing missing values or values outside the scale range
## (`geom_point()`).

But, these are all stacked up… to actually see them, use “geom_jitter” instead of points

penguins %>% ggplot(aes(x = species, y = body_mass_g)) + geom_boxplot() + geom_jitter()

## Warning: Removed 2 rows containing non-finite outside the scale range
## (`stat_boxplot()`).

## Warning: Removed 2 rows containing missing values or values outside the scale range
## (`geom_point()`).

How to get the boxplots on top? The layers are plotted in the order you give them, so change to geom_point + geom_boxplot. (You might want to change the alpha on the boxplot to be able to see the plots under them)

penguins %>% ggplot(aes(x = species, y = body_mass_g)) + geom_jitter() + geom_boxplot(alpha = .7)

## Warning: Removed 2 rows containing non-finite outside the scale range
## (`stat_boxplot()`).

## Warning: Removed 2 rows containing missing values or values outside the scale range
## (`geom_point()`).

Maybe let’s try replacing the boxplot with a geom_violin()?

penguins %>% ggplot(aes(x = species, y = body_mass_g)) + geom_jitter() + geom_violin(alpha = .7)

## Warning: Removed 2 rows containing non-finite outside the scale range
## (`stat_ydensity()`).

## Warning: Removed 2 rows containing missing values or values outside the scale range
## (`geom_point()`).

If time: More Practice with Penguins

• Count the number of male and female penguins for each species.
• Find the average penguin mass for each island
• Find the average bill length by sex
• Make a plot of bill_length and bill_depth, but only for the female penguins.
• Make a plot of flipper length and body mass, faceted by island and colored by species.
• Plot the density of the flipper length, colored by sex
• Plot the density of the body mass, colored by year

penguins %>% group_by(species, sex) %>% summarise(n = n()) %>% pivot_wider(names_from = sex, values_from = n, names_prefix = "n_")

## `summarise()` has grouped output by 'species'. You can override using the
## `.groups` argument.

## # A tibble: 3 × 4
## # Groups:   species [3]
##   species   n_female n_male  n_NA
##   <fct>        <int>  <int> <int>
## 1 Adelie          73     73     6
## 2 Chinstrap       34     34    NA
## 3 Gentoo          58     61     5

penguins %>% group_by(island) %>% summarise(avg_mass = mean(body_mass_g, na.rm=TRUE)) %>% pivot_wider(names_from = island, values_from = avg_mass, names_prefix = "avg_mass_")

## # A tibble: 1 × 3
##   avg_mass_Biscoe avg_mass_Dream avg_mass_Torgersen
##             <dbl>          <dbl>              <dbl>
## 1           4716.          3713.              3706.

penguins %>% group_by(sex) %>% summarise(avg_bill_len = mean(bill_length_mm, na.rm=TRUE)) %>% pivot_wider(names_from = sex, values_from = avg_bill_len, names_prefix = "avg_bill_len_")

## # A tibble: 1 × 3
##   avg_bill_len_female avg_bill_len_male avg_bill_len_NA
##                 <dbl>             <dbl>           <dbl>
## 1                42.1              45.9            41.3

penguins %>% filter(sex == "female") %>% ggplot(aes(x = bill_length_mm, y = bill_depth_mm, color = species)) + geom_point() + labs(title = "Female Penguin Bills", x = 'Bill Length (mm)', y = 'Bill depth (mm)') + theme_bw()

penguins %>% ggplot(aes(x = bill_length_mm, y = body_mass_g, color = species)) + geom_point() + labs(title = "Bill Stats by Island", x = 'Bill Length (mm)', y = 'Body Mass (g)', color = 'Species') + facet_wrap(~ island) + theme_bw()

## Warning: Removed 2 rows containing missing values or values outside the scale range
## (`geom_point()`).

penguins %>% ggplot(aes(x = flipper_length_mm, color = sex)) + geom_density() + labs(title = "Flipper Length by Sex", x = 'Flipper Length (mm)', color = 'Sex') + theme_bw()

## Warning: Removed 2 rows containing non-finite outside the scale range
## (`stat_density()`).

penguins %>% ggplot(aes(x = body_mass_g, color = year)) + geom_density() + labs(title = "Body Mass by Year", x = 'Body Mass (g)', color = 'Year') + theme_bw()

## Warning: Removed 2 rows containing non-finite outside the scale range
## (`stat_density()`).

## Warning: The following aesthetics were dropped during statistical transformation:
## colour.
## ℹ This can happen when ggplot fails to infer the correct grouping structure in
##   the data.
## ℹ Did you forget to specify a `group` aesthetic or to convert a numerical
##   variable into a factor?